U.S. patent application number 14/554891 was filed with the patent office on 2015-05-28 for systems and methods for probability based risk prediction.
The applicant listed for this patent is FedEx Corporation, Washington State University. Invention is credited to Richard A. Lewis, Suresh Rangan, Hans P.A. Van Dongen.
Application Number | 20150148616 14/554891 |
Document ID | / |
Family ID | 53183197 |
Filed Date | 2015-05-28 |
United States Patent
Application |
20150148616 |
Kind Code |
A1 |
Van Dongen; Hans P.A. ; et
al. |
May 28, 2015 |
SYSTEMS AND METHODS FOR PROBABILITY BASED RISK PREDICTION
Abstract
Systems and methods for probability based prediction of risks
are disclosed herein. In one embodiment, a method includes
measuring, using sensing elements individually associated with
multiple users, a first set of data associated with the users. The
method also includes transmitting to and storing in a computing
device the measured first set of data and applying a performance
model to the first set of data to generate a first set of
performance values. The method further includes estimating a second
set of data associated with the one or more users and applying the
performance model to the second set of data to generate a second
set of performance values. The method yet further includes
determining if the first and second sets of performance values are
statistically equivalent to each other.
Inventors: |
Van Dongen; Hans P.A.;
(Spokane, WA) ; Rangan; Suresh; (Germantown,
TN) ; Lewis; Richard A.; (Olive Branch, MS) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Washington State University
FedEx Corporation |
Pullman
Memphis |
WA
TN |
US
US |
|
|
Family ID: |
53183197 |
Appl. No.: |
14/554891 |
Filed: |
November 26, 2014 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61909826 |
Nov 27, 2013 |
|
|
|
Current U.S.
Class: |
600/300 |
Current CPC
Class: |
G16H 40/63 20180101;
G16H 50/20 20180101; A61B 5/7275 20130101; A61B 5/4806 20130101;
G16H 50/30 20180101; G16H 40/67 20180101; A61B 5/18 20130101 |
Class at
Publication: |
600/300 |
International
Class: |
A61B 5/00 20060101
A61B005/00; A61B 5/18 20060101 A61B005/18 |
Claims
1. A method, comprising: measuring, using sensing elements
individually associated with multiple users, a first set of sleep
data associated with the users operating according to a first duty
schedule, the sleep data containing at least one or more of sleep
starting time, sleep ending time, or time of day during sleep
corresponding to each user; transmitting to and storing in a
computing device the measured first set of sleep data; with the
computing device, applying, a fatigue model to the first set of
sleep data to generate a first set of fatigue values; estimating a
second set of sleep data associated with the users according to a
second duty schedule different than the first duty schedule without
subjecting the individuals to perform duties according to the
second duty schedule; applying the fatigue model to the second set
of sleep data to generate a second set of fatigue values; and
determining if the second duty schedule is equivalent to the first
duty schedule based on the generated first and second sets of
fatigue values.
2. The method of claim 1, further comprising, in response to
determining that the second duty schedule is not equivalent or
non-inferior to the first duty schedule, estimating a third set of
sleep data associated with the one or more individuals according to
a third duty schedule different than the first and second duty
schedules without subjecting the individuals to perform duties
according to the second duty schedule, applying the fatigue model
to the third set of sleep data to generate a third set of fatigue
values, and determining if the third duty schedule is statistically
equivalent to the first duty schedule based on the generated first
and third sets of fatigue values.
3. The method of claim 1 wherein both the first and second sets of
fatigue values include a distribution of values, and wherein the
process performed by the processor further includes calculating one
or more statistical parameters for each of the first and second
sets of fatigue values.
4. The method of claim 1 wherein: both the first and second sets of
fatigue values include a distribution of values; the process
performed by the processor further includes calculating one or more
statistical parameters for each of the first and second sets of
fatigue values; and determining if the second duty schedule is
statistically equivalent or non-inferior to the first duty schedule
includes determining if the second duty schedule is statistically
equivalent or non-inferior to the first duty schedule based on the
calculated one or more statistical parameters for each of the first
and second sets of fatigue values.
5. A computing system having a processor and a memory containing
instructions that when executed by the processor, cause the
processor to perform a process comprising: applying a fatigue model
on a first set of sleep data associated with one or more
individuals according to a first duty schedule to generate a first
set of fatigue values, the sleep data containing at least one or
more of sleep starting time, sleep ending time, or time of day
during sleep corresponding to each individual; estimating a second
set of sleep data associated with the one or more individuals
according to a second duty schedule different than the first duty
schedule without subjecting the individuals to perform duties
according to the second duty schedule; applying the fatigue model
to the second set of sleep data to generate a second set of fatigue
values; and determining if the second duty schedule is
statistically equivalent to the first duty schedule based on the
generated first and second sets of fatigue values.
6. The computing system of claim 5 wherein the process performed by
the processor further includes, in response to determining that the
second duty schedule is not statistically equivalent to the first
duty schedule, estimating a third set of sleep data associated with
the one or more individuals according to a third duty schedule
different than the first and second duty schedules without
subjecting the individuals to perform duties according to the
second duty schedule, applying the fatigue model to the third set
of sleep data to generate a third set of fatigue values, and
determining if the third duty schedule is statistically equivalent
to the first duty schedule based on the generated first and third
sets of fatigue values.
7. The computing system of claim 5 wherein both the first and
second sets of fatigue values include a distribution of values, and
wherein the process performed by the processor further includes
calculating one or more statistical parameters for each of the
first and second sets of fatigue values.
8. The computing system of claim 5 wherein: both the first and
second sets of fatigue values include a distribution of values; the
process performed by the processor further includes calculating one
or more statistical parameters for each of the first and second
sets of fatigue values; and determining if the second duty schedule
is statistically equivalent to the first duty schedule includes
determining if the second duty schedule is statistically equivalent
to the first duty schedule based on the calculated one or more
statistical parameters for each of the first and second sets of
fatigue values.
9. The computing system of claim 5 wherein: both the first and
second sets of fatigue values include a distribution of values; the
process performed by the processor further includes calculating a
statistical parameter for each of the first and second sets of
fatigue values, the statistical parameter including one of a mean,
an average, a range, one or more quantile values, a standard
deviation, a mean difference, a median absolute deviation, an
average absolute deviation, or a distance standard deviation; and
determining if the second duty schedule is statistically equivalent
to the first duty schedule includes determining if the second duty
schedule is statistically equivalent to the first duty schedule
based on the calculated statistical parameter for each of the first
and second sets of fatigue values.
10. The computing system of claim 5 wherein: both the first and
second sets of fatigue values include a distribution of values; the
process performed by the processor further includes calculating a
statistical parameter for each of the first and second sets of
fatigue values, the statistical parameter including one of a mean,
an average, a range, one or more quantile values, a standard
deviation, a mean difference, a median absolute deviation, an
average absolute deviation, or a distance standard deviation;
determining if the second duty schedule is statistically equivalent
to the first duty schedule includes determining if the second duty
schedule is statistically equivalent to the first duty schedule
based on the calculated statistical parameter for each of the first
and second sets of fatigue values; and the process performed by the
processor further includes indicating that the second duty schedule
is statistically equivalent to the first duty schedule in response
to determining that a difference between the calculated statistical
parameters is within a threshold.
11. A method performed by a computing device having a processor,
comprising: with the processor, receiving a first set of sleep data
associated with one or more individuals operating according to a
first duty schedule, the sleep data containing at least one or more
of sleep starting time, sleep ending time, or time of day during
sleep corresponding to each individual; receiving a second set of
sleep data associated with the one or more individuals operating
according to a second duty schedule different than the first duty
schedule; generating first and second sets of fatigue values based
on the first and second sets of sleep data, respectively; and
determining a statistical relationship of fatigue risk between the
first duty schedule and the second duty schedule based on the
generated first and second sets of fatigue values.
12. The method of claim 11 wherein at least one of receiving the
first set of sleep data or receiving the second set of sleep data
includes receiving the first set of sleep data from a rest/activity
monitor associated with each individual on a periodic or continuous
basis.
13. The method of claim 11 wherein receiving the second set of
sleep data includes estimating the second set of sleep data based
at least on a circadian parameter corresponding to each individual
or estimating the second set of sleep data by recording the sleep
data of the individuals without the individuals actually performing
duties.
14. The method of claim 11 wherein receiving the first set of sleep
data and receiving the second set of sleep data include estimating
the first set and second set of the sleep data, respectively.
15. The method of claim 11 wherein receiving the second set of
sleep data includes estimating the second set of sleep data by
obtaining sleep data from areas of operation other than that
associated with the individuals or by interpolating, extrapolating,
or imputing from the first set of sleep data.
16. The method of claim 11 wherein at least one of receiving the
first set of sleep data or receiving the second set of sleep data
includes sampling from a set of sleep data associated with the
first or second schedule, respectively.
17. The method of claim 11 wherein determining the statistical
relationship includes calculating a statistical parameter for each
of the generated first and second sets of fatigue values, the
statistical parameter including one of a mean, an average, a range,
one or more quantile values, a standard deviation, a mean
difference, a median absolute deviation, an average absolute
deviation, or a distance standard deviation.
18. The method of claim 11 wherein determining the statistical
relationship includes calculating a statistical parameter for each
of the generated first and second sets of fatigue values and
comparing the calculated statistical parameters corresponding to
the generated first and second sets of fatigue values.
19. The method of claim 11 wherein: determining the statistical
relationship includes calculating a statistical parameter for each
of the generated first and second sets of fatigue values, the
statistical parameter including one of a mean, an average, a range,
one or more quantile values, a standard deviation, a mean
difference, a median absolute deviation, an average absolute
deviation, or a distance standard deviation; determining the
statistical relationship further includes: comparing the calculated
statistical parameters corresponding to the generated first and
second sets of fatigue values; and indicating that the second duty
schedule is statistically non-inferior if the calculated
statistical parameter of the generated second set of fatigue values
is greater than or equal to that of the first set of fatigue
values.
20. The method of claim 11 wherein: determining the statistical
relationship includes calculating a statistical parameter for each
of the generated first and second sets of fatigue values, the
statistical parameter including one of a mean, an average, a range,
one or more quantile values, a standard deviation, a mean
difference, a median absolute deviation, an average absolute
deviation, or a distance standard deviation; determining the
statistical relationship further includes: comparing the calculated
statistical parameters corresponding to the generated first and
second sets of fatigue values; in response to that the calculated
statistical parameter of the generated second set of fatigue values
is greater than or equal to that of the first set of fatigue
values, indicating that the second duty schedule is statistically
non-inferior than the first duty schedule; and in response to that
the calculated statistical parameter of the generated second set of
fatigue values is less than that of the first set of fatigue
values, estimating a third set of sleep data associated with the
one or more individuals operating according to a third duty
schedule different than the first or second duty schedule; and
repeating the generating and determining operations.
Description
CROSS-REFERENCE TO RELATED APPLICATION(S)
[0001] This application claims priority to U.S. Provisional
Application No. 61/909,826, filed on Nov. 27, 2013, the disclosure
of which is incorporated herein by reference in its entirety.
BACKGROUND
[0002] Fatigue from sleep loss, circadian misalignment, time on
task, or other sources can degrade cognitive functioning and
impairs performance, productivity, and safety. The personal,
economic, and social costs involved in errors, incidents, and
accidents resulting from fatigue are considerable in commercial
aviation, trucking, mining, and other areas.
[0003] In Jan. 4, 2012, the Federal Aviation Administration ("FAA")
published flight and duty regulations on pilot scheduling for
commercial flights to ensure pilots have adequate opportunity to
rest before each duty day. The rules also allow airlines to develop
alternative ways of mitigating pilot fatigue (i.e., alternative
means of compliance) based on science and validated by sleep,
performance, and safety data submitted to the FAA for approval.
However, the rules create dilemmas for airlines attempting to
develop an alternative means of compliance. On one hand, airlines
cannot show compliance without collecting fatigue data when pilots
operated according to the alternative means of compliance. On the
other, airlines are not allowed to operate according to the
alternative means of compliance without having the FAA's approval
first based on the data.
BRIEF DESCRIPTION OF THE DRAWINGS
[0004] FIG. 1 is a sleep homeostatic pressure versus time graph
that shows an example study for fatigue distribution modeling in
accordance with embodiments of the disclosed technology.
[0005] FIG. 2 is a schematic block diagram illustrating a fatigue
prediction system in accordance with additional embodiments of the
disclosed technology.
[0006] FIG. 3 is a flow diagram illustrating a process for
probability based prediction of fatigue risk in accordance with
embodiments of the disclosed technology.
[0007] FIG. 4 is a computing device suitable for certain components
of the fatigue prediction system in FIG. 2.
DETAILED DESCRIPTION
[0008] The present disclosure describes systems and methods for
probability based prediction of fatigue risks. Based on such
prediction, alternative means of compliance with scheduling rules
promulgated by the FAA or other regulatory entities may be readily
developed. Embodiments of the disclosed technology may also be used
to compare fatigue distributions of two or more scheduling options
for other reasons than (or in addition to) extant regulations. Even
though particular systems, devices, components, and operations are
disclosed in the following description, a person skilled in the
relevant art will also understand that the technology may have
additional embodiments, and that the technology may be practiced
without several of the details of the embodiments described below
with reference to FIGS. 1-4.
[0009] Fatigue resulting from sleep loss and circadian rhythm can
be associated with decreased alertness reflected in reduced
capacity to perform cognitive tasks and increased variability in
performance. Thus, greater possibility of errors, incidents and
accidents may result when pilots, drivers, or operators are
fatigued. In around-the-clock operations, neurobiological
mechanisms underlying fatigue mainly involve a sleep/wake
homeostatic process and a circadian process. The sleep/wake
homeostatic process tracks sleep history and seeks to balance time
spent awake with an appropriate amount of recuperative sleep. The
circadian process, driven by the biological clock in the brain,
tracks time of day and seeks to place wakefulness during the day
and sleep during the night.
[0010] The homeostatic and circadian processes normally operate in
tandem to provide a stable level of alertness during the day and
consolidated sleep during the night. However, when deviating from a
normal schedule of daytime wakefulness and nighttime sleep,
interaction of the two processes can lead to fatigue and associated
decreased alertness and performance impairment. During nighttime
operations, for example, the homeostatic and circadian processes
align to steadily increase fatigue overnight, while also leading to
difficulty to sleep during the day. When crossing time zones, the
circadian process becomes temporarily desynchronized and may take
several days or even weeks to adjust to the new time zone.
[0011] As used herein, the term "fatigue" generally refers to a
level of deterioration in a person's alertness (as reflected in
performance capability) as a function of sleep/wake history (time
awake), circadian rhythm (time of day), sleep inertia (transient
sleepiness immediately after awakening), workload (time on task,
duty hours, nature of work), and/or other suitable factors. The
experimentally determined effects of sleep/wake history and
circadian rhythm on sleep propensity, alertness, and performance
may be used to develop a mathematical model (referred herein as a
"fatigue model") for predicting fatigue based on such factors. One
example can include a two-process model that invokes the
homeostatic drive for sleep and the circadian rhythm in sleep
propensity as processes driving sleepiness and fatigue. Other
mathematical models can also use shift timing and duration
(constituting a rough estimate of workload and/or time awake), as
well as time of day (constituting a rough estimate of circadian
rhythm phase) as their inputs.
[0012] With a fatigue model, it is believed that the impact of
fatigue on a duty schedule can be evaluated without having to test
that schedule in actual operation. As used herein, a "duty
schedule" generally refers to a schedule according to which one or
more individuals perform designated tasks, for example, flying a
plane, driving a truck, operating machinery, etc. A fatigue model
can be validated on specific data sets, and may then be generalized
to predict the performance consequences of a potential schedule.
The predictions of a fatigue model can be adjustable to predict
objectively measurable loss in productivity (e.g., increased fuel
consumption and/or increased maintenance in transportation) or
other operationally relevant performance outcomes. One suitable
fatigue model is disclosed in a Publication by Gregory D. Roach et
al., entitled "A MODEL TO PREDICT WORK-RELATED FATIGUE BASED ON
HOURS OF WORK," published in Aviation, Space, and Environmental
Medicine, Vol. 75, No. 3, Section II, March 2004, the disclosure of
which is incorporated herein in its entirety.
[0013] The inventors have recognized that sleep timing and duration
not only depend on biological processes (e.g., homeostatic and
circadian processes) but also on non-biological factors, especially
when individuals work consecutive nights and/or traversing multiple
time zones. Aside from the sleep/wake/work schedule itself, these
factors range from availability of hotel facilities (check-in/out
times) or store opening hours to communications with family at
home, etc. Some of these factors vary substantially from person to
person and from duty to duty, such that pre-duty, in-flight,
layover, and post-duty sleep schedules exhibit probabilistic
distributions. A sleep prediction model for sleep timing and
duration can be developed based not just on the underlying
biological processes but also on systematic patterns in
observations of real-world in-flight and layover sleep behaviors,
as a function of prior and planned duty schedule, time of day,
and/or location. The development of the sleep prediction model
allows for fatigue distribution modeling by making probabilistic
predictions of alertness that account for components of natural
variability in sleep behavior.
[0014] FIG. 1 is a sleep homeostatic pressure versus time graph
that shows an example study for fatigue distribution modeling in
accordance with embodiments of the disclosed technology. Data shown
in FIG. 1 were obtained from the study using wrist-worn
rest/activity monitors in eleven pilots for a simulated forty eight
hour duty schedule. The schedule involved a nine hour duty period
starting at 01:00 AM followed by a twenty five hour layover period.
A second twelve hour duty period starting at noon followed the
layover period. Alertness predictions were made across the forty
eight hour scenario using a two-process model. FIG. 1 only shows
the homeostatic sleep pressure component of the alertness
predictions. The circadian rhythm component of the alertness
predictions is not shown in FIG. 1 for clarity purposes.
[0015] As shown in FIG. 1, a first duty period 102 is followed by a
layover period 104 with sleep availability. A second duty period
106 then follows the layover period 104. During the first and
second duty periods 102 and 106, the individuals had duty schedules
112 and 114 that have varying duty periods. In the illustrated
embodiment, the duty periods in each of the duty schedules 112 and
114 overlap with one another. In other embodiments, at least one of
the duty periods in each of the duty schedules 112 and 114 may not
overlap with others.
[0016] FIG. 1 also shows homeostatic sleep pressure values of the
individuals as a function of time observed during the study. Each
line in FIG. 1 represents a homeostatic value for one individual.
The homeostatic value increases when an individual was awake and
decreases when the individual was asleep. For example, as shown in
FIG. 1, an individual 100 had a homeostatic value that increases
during the first duty period 102 until a time point 101a when the
layover period 104 starts. The homeostatic value of the individual
100 then decreased from the time point 101a to a time point 101b by
sleeping during this period. The individual 100 then woke up after
the time point 101b and stayed awake until a time point 101c when
the individual 100 went back to sleep again until a time point
101d. Even though the individual 100 could sleep from the time
point 101a to the time point 101d, the individual 100 had two
sleeping periods (i.e., between the first and second time points
101a and 101b and between the third and fourth time points 101c and
101d) before the second duty period 106 starts. The other
individuals also exhibited similar behaviors (but not exactly the
same) as reflected in the two distributed sleep patterns 116a and
116b.
[0017] Based on the observed sleep pattern distribution, fatigue
(or a reduction of alertness) distribution 118 can be predicted for
the second duty period 106. In the study, at the end of the second
duty period 106, alertness scores of the pilots were 0.35.+-.0.05
(mean.+-.standard deviation). Forty eight hours earlier, the
alertness score of the same pilots was 0.67. Such alertness
distribution information can be used to statistically compare the
predictions to those of other possible duty schedules, and to make
informed decisions regarding the need for alertness-enhancing
countermeasures.
[0018] Several embodiments of the disclosed technology are directed
to probability based prediction of fatigue risks utilizing a
fatigue model. In certain embodiments, a first set of data
representing sleep patterns of one or more individuals (e.g.,
pilots, drivers, operator, etc.) operating under a first duty
schedule are measured. A second set of data representing sleep
patterns of the individuals operating under a second duty schedule
can be estimated based on biology, experiments, measured data from
other operation areas, and/or other suitable information. The
fatigue model can then be applied to both the first and second sets
of data representing sleep patterns to generate first and second
distributions of fatigue levels (or alertness) of the individuals.
Statistical analysis can then be applied to the first and second
distributions of fatigue levels (or alertness) to measure a
probability of equivalency (or non-inferiority) between the first
and second duty schedules. As such, airlines, trucking companies,
or other suitable entities may develop alternative means of
compliance with governmental scheduling rules without the need to
actually operate under proposed alternative schedules, or may
develop alternative schedules that are less fatiguing, more
productive, and/or safer than existing schedules.
[0019] FIG. 2 is schematic block diagrams illustrating probability
based prediction of fatigue risks in accordance with embodiments of
the disclosed technology. In FIG. 2 and in other Figures herein,
individual software modules, components, and routines may be a
computer program, procedure, or process written as source code in
C, C#, C++, Java, and/or other suitable programming languages. The
computer programs, procedures, or processes may be compiled into
intermediate, object, or machine code and presented for execution
by a processor of a personal computer, a network server, a laptop
computer, a server computer, or other suitable computing devices.
Various implementations of the source, intermediate, and/or object
code and associated data may be stored in one or more computer
readable storage media that include read-only memory, random-access
memory, magnetic disk storage media, optical storage media, flash
memory devices, and/or other suitable media. As used herein, the
term "computer readable storage media" excludes propagated signals,
per se.
[0020] As shown in FIG. 2, the fatigue prediction system can
include a computing processor 202 operatively coupled to a database
204. The computing processor 202 can include a micro-processor, a
field programmable gate array, and/or other suitable processing
elements. The database 204 can contain records of measured sleep
data 232, estimated sleep data 234, and one or more fatigue models
236. Examples of the measured and/or estimated sleep data 232 and
234 can include at least one or more of sleep starting time, sleep
ending time, time of day during sleep, and/or other suitable
information. In certain embodiments, the individuals 100 can have a
generally similar starting state or alertness level, for example,
as shown in FIG. 1. In other embodiments, the individuals 100 can
have distributed starting states. In the illustrated embodiment,
the computing processor 202 and the database 204 are shown as being
integrated into the fatigue prediction system 200. In other
embodiments, the database 204 may be hosted by a remote server or a
plurality of remote servers (not shown).
[0021] The computing processor 202 can be configured to execute
instructions contained in a memory (not shown in FIG. 2) that
provide a measurement module 222, an estimation module 224, a
fatigue module, and an analysis module 228. Even though particular
modules are show in FIG. 2, in other embodiments, the computing
processor 202 may also include input modules (e.g., keyboard
drivers), output modules (e.g., printer drivers), communications
modules (e.g., network drivers), and/or other suitable types of
modules. In further embodiments, various modules of the computing
processor 202 may also be executed by additional and/or different
computing processors (not shown).
[0022] The measurement module 222 is configured to obtain the
measured sleep data 232 from one or more individuals 100 (e.g.,
pilots, drivers, machinery operators, etc.) operating according to
a base duty schedule. In one embodiment, the base duty schedule can
be a schedule approved by a governmental entity (e.g., FAA). In
other embodiments, the base duty schedule can be an existing
operating schedule or other suitable types of schedule. In the
illustrated embodiment, each individual 100 carries or is otherwise
associated with a sensing element 206 configured to detect and
record measured sleep data 232. In one example, the sensing element
206 can include a portable rest/activity monitor that can be
carried by the individuals 100. One suitable monitor (model No.
GT9X ActiGraph Link) is provided by the ActiGraph Company of
Pensacola, Fla. In other examples, the sensing element 206 can
include a wrist actigraphy device, a fab, a smartphone, a laptop
computer, and/or other suitable devices configured to record a
sleep log from the individuals 100.
[0023] In certain embodiments, the sensing elements 206 may upload
the recorded measured sleep data 232 to the measurement module 222
on a periodic, continuous, or other suitable basis via a computer
network, a hardwire connection, a mobile phone network, and/or
other suitable types of communications channel 238. In other
embodiments, the measurement module 222 may poll the various
sensing elements 206 on a periodic or other suitable basis. In yet
other embodiments, the measurement module 222 can prompt the
individuals 100 or other suitable operators to download the
recorded measured sleep data 232 on a periodic or other suitable
basis. In further embodiments, the sensing elements 206 may be
omitted, and the individuals 100 may supply the measured sleep data
232 to the measurement module 222 by logging into a website (not
shown) linked to the measurement module 222 and input the sleep
data 232. In yet further embodiments, the measurement module 222
may also obtain the measured sleep data 232 using implantable
devices (not shown), manual recording, or other suitable components
and/or techniques.
[0024] The measurement module 222 is also configured to organize
and store the measured sleep data 232 in the database 204. In one
embodiment, the measurement module 222 can be configured to create
records of the measured sleep data 232 corresponding to each of the
individuals 100 based on at least one of an identification of the
individual 100, a group the individual 100 belongs to, a work
schedule, and/or other suitable criteria. In other embodiments, the
measurement module 222 can also be configured to calculate certain
sleep parameters based on the measured sleep data 232. For example,
the measurement module 222 can be configured to calculate a
duration of sleep, an average duration of sleep over a period of
time, and/or other suitable parameters. In further embodiments, the
measurement module 222 can be configured to manipulate and/or
otherwise process the measured sleep data 232 before storing the
measured sleep data 232 in the database 204.
[0025] The estimation module 224 is configured to generate a set of
estimated sleep data 234 for the individuals 100 based on a test
duty schedule different than the base duty schedule. In one
embodiment, the test duty schedule can be a proposed duty schedule
as an alternative means of compliance with a governmental
regulation. In other embodiments, the test duty schedule may be a
schedule under study or other suitable types of schedule. In
certain embodiments, the estimation module 224 can be configured to
obtain the estimated sleep data 234 based on circadian parameters,
age, gender, and/or other suitable biological parameters of the
individuals 100. In another embodiments, the estimation module 224
can generate the estimated sleep data 234 by observing and
recording sleep data of the individuals 100 in an experimental
setting without the individuals 100 actually performing the duties
according to the test duty schedule.
[0026] In yet another embodiment, the estimation module 224 can
generate the estimated sleep data 234 from areas of operation other
than that associated with the test duty schedule and/or the
individuals 100. For example, the estimated sleep data 234 for
pilots may be generated based on sleep data for ground crews who
operate according to a schedule generally similar to or the same as
the test duty schedule. In a further embodiment, the estimation
module 224 can also estimate the set of estimated sleep data 234
based on the measured sleep data 232 via interpolation,
extrapolation, and/or other suitable techniques. In yet further
embodiments, the estimated sleep data 234 can also be obtained from
individuals 100 actually operating according to the test duty
schedule, for example, because an exemption has been granted to
collect data, or because the regulations are not yet applicable
(e.g., before an effective date of the regulations). In the
illustrated embodiment, the estimated sleep data 234 are stored in
the database 204. In other embodiments, the estimated sleep data
234 may be stored in other suitable locations.
[0027] The fatigue module 226 is configured to apply the fatigue
model 236 to the measured sleep data 232 and the estimated sleep
data 234 to generate (1) a set of fatigue values corresponding to
the base duty schedule (referred to as "base fatigue values"); and
(2) a set of fatigue values corresponding to the test duty schedule
(referred to as "test fatigue values"). In certain embodiments, the
fatigue module 226 can be applied to the measured and estimated
sleep data 232 and 234 on an individual basis. In other
embodiments, the fatigue module 226 can be applied to permutations
of the measured or estimated sleep data 232 and 234 by combining
portions of the sleep data 232 and 234. For instance, one
permutation of the measured sleep data 232 can include the first
sleep period between the time points 101a and 101b of the first
individual 100 and a second sleep period approximately between the
time points 101c and 101d of a second individual 100 in FIG. 1. In
such embodiments, random or non-random sampling from possible
permutations may be used to limit a computational burden if a
number of the permutations escalates. As discussed above, the
measured sleep data 232 and the estimated sleep data 234 both
include a distribution of values. As a result, the base fatigue
values and the test fatigue values can both include a distribution
of values after the fatigue model 236 is applied. As discussed in
more detail below, the base and test fatigue values may then be
analyzed by the analysis module 228 to determine if the test duty
schedule carries a fatigue risk higher than, generally equivalent
to, or lower than the base duty schedule.
[0028] The analysis module 228 is configured to apply statistical
analysis or other suitable comparison technique (e.g. graphical,
tabular, and/or visual comparison of distribution overlap) on the
base and test fatigue values and can include various routines
configured to calculate various statistical parameters. For
example, in one embodiment, the analysis module 228 is configured
to determine a mean, average, range, variability (e.g., standard
deviation, interquartile range, median absolute deviation, etc.),
or other suitable statistical parameters of the base and test
fatigue values. The analysis module 228 can then compare the
determined statistical parameters of the base and test fatigue
values. In one example, the analysis module 228 can indicate that
the base duty schedule carries a lower fatigue risk if at least one
of the following statistical parameters associated with the test
fatigue values are lower than corresponding ones associated with
the base fatigue values: [0029] mean fatigue value; [0030] average
fatigue value; [0031] range of fatigue value; [0032] standard
deviation of fatigue values; [0033] mean difference of fatigue
values; [0034] median absolute deviation of fatigue values; [0035]
average absolute deviation of fatigue values; [0036] distance
standard deviation of fatigue values. In certain embodiments, the
foregoing comparison may be performed across the entire span of the
base and test duty schedules, across selected portions of the
foregoing duty schedules, or as associated with specific sections
or events such as critical phases of a flight (e.g., take-offs and
landings). In other embodiments, equivalence testing may be
performed only on quantiles of the distributions that represent the
greatest risk (e.g., the right-hand tails of the distributions).
Such an approach may focus on the risks that may lead to the most
catastrophic outcome rather than risks in general. In other
examples, the foregoing determination may be based on other
suitable statistical parameters or test statistics (e.g., t test
statistics, F statistics, chi-square, etc.).
[0037] In operation, the measurement module 222 obtains the
measured sleep data 232 from, for example, the sensing elements 206
associated with the individuals 100. The estimation module 224 can
estimate or otherwise generate the estimated sleep data 234 based
on the test duty schedule. In one embodiment, the test duty
schedule can be input by an operator (not shown). In another
embodiment, the test duty schedule can be generated by an
application or process executed by the computing processor 202 or
other processors (not shown). In yet further embodiments, the test
duty schedule may be generated in other suitable manners.
[0038] The fatigue module 226 can then apply the fatigue model 236
to both the measured sleep data 232 and the estimated sleep data
234 to generate the base and test fatigue values, respectively. The
analysis module 228 can then perform statistical analysis on the
base and test fatigue values to determine if the test fatigue
values are statistically superior than, equivalent to, or inferior
than the base fatigue values. In response to determining that the
test fatigue values are statistically superior than and/or
equivalent to the base fatigue values, the analysis module 228 can
indicate that the test duty schedule is at least equivalent, if not
superior to, the base duty schedule.
[0039] In accordance with certain aspects of the disclosed
technology, the test fatigue values are not generated based on data
measured when the individuals 100 actually perform duties according
to the test duty schedule. Instead, the test fatigue values are
generated based on the estimated sleep data 232 without the need to
actually having the individuals perform duties according to the
test duty schedule. Thus, airlines may develop alternative means of
compliance with the FAA's scheduling rules even without the FAA's
prior approval for the test duty schedule for the individuals 100.
As a result, sufficient savings in operating expenses as well as
greater safety to pilots and the public may be achieved. In further
embodiments, other suitable information (e.g., age, rank,
experience, and/or genotype) may also be used to affect the
predictions. The suitable information may also include actual
fatigue data obtained during part of the schedule (e.g., the
beginning part that may be flown without exceeding the applicable
regulations). Such data may replace portions of the test fatigue
values from the fatigue model predictions where applicable (e.g.,
to generate initial values) or be combined therewith using the
Bayesian technique as disclosed in U.S. Pat. No. 8,781,796, the
disclosure of which is incorporated herein in its entirety.
[0040] FIG. 3 is a flow diagram illustrating a process 300 for
probability based prediction of fatigue risk in accordance with
embodiments of the disclosed technology. As shown in FIG. 3, the
process 300 includes obtaining measured sleep data of individuals
operating according to a base duty schedule at stage 302. As
discussed in more detail above with reference to FIG. 2, the
measured sleep data may be obtained via the sensing elements 206
(FIG. 2) or via other suitable components. The process 300 can then
include applying a fatigue model to the measured sleep data 304 to
generate a set of base fatigue values at stage 304. In one
embodiment, the fatigue model can be represented as a function of
homeostatic and circadian parameters. Applying the fatigue model
includes calculating a set of base fatigue values using the
measured sleep data as input to the function. In other embodiments,
the fatigue model may be represented as tables, graphs, and/or in
other suitable manners. Applying the fatigue model can include
table or graph lookups and/or other suitable operations.
[0041] The process 300 can also include obtaining estimated sleep
data related to a test duty schedule for the individuals at stage
306. As discussed above with reference to FIG. 2, the estimated
sleep data may be obtained in various fashions without the need to
actually measure sleep data of the individual operating according
to the test duty schedule. The process 300 can then include
applying the same fatigue model to the estimated sleep data at
stage 308 to generate a set of test fatigue values.
[0042] The process 300 can further include comparing the set of
base fatigue values to the set of test fatigue values to determine
a statistical relationship between the two sets of data at stage
310. Various suitable techniques for performing such statistical
analysis are discussed above with reference to the analysis module
228 in FIG. 2. The process 300 can then include a decision stage
312 to determine if the test duty schedule is better than the base
duty schedule based on the comparison of the base and test fatigue
values. In one embodiment, the test duty schedule is indicated to
be better than the base duty schedule if the test fatigue values
are statistically equivalent or superior than the base fatigue
values. In one example, the test duty schedule is statistically
equivalent to the base duty schedule if a difference between
corresponding statistical parameters (e.g., a mean value) of the
base and test fatigue values is within a predetermined threshold.
In other examples, the foregoing determination can also be based on
other suitable criteria. In response to determining that the test
duty schedule is better than the base duty schedule, the process
300 can include outputting the test duty schedule at stage 314;
otherwise, the process 300 can revert to stage 306 based on another
test duty schedule.
[0043] FIG. 4 is a computing device 400 suitable for the fatigue
prediction system in FIG. 2. In a very basic configuration 402,
computing device 400 typically includes one or more processors 404
and a system memory 406. A memory bus 408 may be used for
communicating between processor 404 and system memory 406.
[0044] Depending on the desired configuration, the processor 404
may be of any type including but not limited to a microprocessor
(.mu.P), a microcontroller (.mu.C), a digital signal processor
(DSP), or any combination thereof. The processor 404 may include
one more levels of caching, such as a level one cache 410 and a
level two cache 412, a processor core 414, and registers 416. An
example processor core 414 may include an arithmetic logic unit
(ALU), a floating point unit (FPU), a digital signal processing
core (DSP Coe), or any combination thereof. An example memory
controller 418 may also be used with processor 404, or in some
implementations memory controller 418 may be an internal part of
processor 404.
[0045] Depending on the desired configuration, the system memory
406 may be of any type including but not limited to volatile memory
(such as RAM), non-volatile memory (such as ROM, flash memory,
etc.) or any combination thereof. The system memory 406 may include
an operating system 420, one or more applications 422, and program
data 424. This described basic configuration 402 is illustrated in
FIG. 4 by those components within the inner dashed line.
[0046] The computing device 400 may have additional features or
functionality, and additional interfaces to facilitate
communications between basic configuration 402 and any other
devices and interfaces. For example, a bus/interface controller 430
may be used to facilitate communications between the basic
configuration 402 and one or more data storage devices 432 via a
storage interface bus 434. The data storage devices 432 may be
removable storage devices 436, non-removable storage devices 438,
or a combination thereof. Examples of removable storage and
non-removable storage devices include magnetic disk devices such as
flexible disk drives and hard-disk drives (HDD), optical disk
drives such as compact disk (CD) drives or digital versatile disk
(DVD) drives, solid state drives (SSD), and tape drives to name a
few. Example computer storage media may include volatile and
nonvolatile, removable and non-removable media implemented in any
method or technology for storage of information, such as computer
readable instructions, data structures, program modules, or other
data.
[0047] The system memory 406, removable storage devices 436 and
non-removable storage devices 438 are examples of computer storage
media. Computer storage media includes, but is not limited to, RAM,
ROM, EEPROM, flash memory or other memory technology, CD-ROM,
digital versatile disks (DVD) or other optical storage, magnetic
cassettes, magnetic tape, magnetic disk storage or other magnetic
storage devices, or any other medium which may be used to store the
desired information and which may be accessed by computing device
400. Any such computer storage media may be part of computing
device 400. The term "computer storage medium" excludes propagated
signals and communication media.
[0048] The computing device 400 may also include an interface bus
440 for facilitating communication from various interface devices
(e.g., output devices 442, peripheral interfaces 444, and
communication devices 446) to the basic configuration 402 via
bus/interface controller 430. Example output devices 442 include a
graphics processing unit 448 and an audio processing unit 450,
which may be configured to communicate to various external devices
such as a display or speakers via one or more A/V ports 452.
Example peripheral interfaces 444 include a serial interface
controller 454 or a parallel interface controller 456, which may be
configured to communicate with external devices such as input
devices (e.g., the sensing elements 206 in FIG. 2, keyboard, mouse,
pen, voice input device, touch input device, etc.) or other
peripheral devices (e.g., printer, scanner, etc.) via one or more
I/O ports 458. An example communication device 446 includes a
network controller 460, which may be arranged to facilitate
communications with one or more other computing devices 462 over a
network communication link via one or more communication ports
464.
[0049] The network communication link may be one example of
communication media. Communication media may typically be embodied
by computer readable instructions, data structures, program
modules, or other data in a modulated data signal, such as a
carrier wave or other transport mechanism, and may include any
information delivery media. A "modulated data signal" may be a
signal that has one or more of its characteristics set or changed
in such a manner as to encode information in the signal. By way of
example, and not limitation, communication media may include wired
media such as a wired network or direct-wired connection, and
wireless media such as acoustic, radio frequency (RF), microwave,
infrared (IR) and other wireless media. The term computer readable
media as used herein may include both storage media and
communication media.
[0050] The computing device 400 may be implemented as a portion of
a small-form factor portable (or mobile) electronic device such as
a cell phone, a personal data assistant (PDA), a personal media
player device, a wireless web-watch device, a personal headset
device, an application specific device, or a hybrid device that
include any of the above functions. The computing device 400 may
also be implemented as a personal computer including both laptop
computer and non-laptop computer configurations.
[0051] From the foregoing, it will be appreciated that specific
embodiments of the disclosed technology have been described herein
for purposes of illustration, but that various modifications may be
made without deviating from the disclosure. Certain aspects of the
disclosure described in the context of particular embodiments may
be combined or eliminated in other embodiments. Not all embodiments
need necessarily exhibit such advantages to fall within the scope
of the disclosure. Accordingly, the invention is not limited by the
disclosure, but instead its scope is to be determined entirely by
the following claims.
* * * * *